add tokenizer

runs/Jun19_16-21-33_container-9db311a83c-fb92563a/1655628870.5865257/events.out.tfevents.1655628870.container-9db311a83c-fb92563a.364.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:019d3a12b9d7e29db4cd5f68bab4a2db2d541ef04934cfb8606aabd4b37c00b4
+size 5380

runs/Jun19_16-21-33_container-9db311a83c-fb92563a/events.out.tfevents.1655628870.container-9db311a83c-fb92563a.364.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e55ba609b10f94a5cedfa41a808000924a9cbd474aa1f40f672b7a86ced81e7
+size 5335

vocab.json

@@ -1,37 +1,37 @@
 {
-  "'": 10,
-  "2": 3,
-  "3": 1,
-  "4": 25,
-  "5": 7,
-  "6": 27,
+  "'": 14,
+  "2": 24,
+  "3": 20,
+  "4": 29,
+  "5": 21,
+  "6": 6,
   "[PAD]": 34,
   "[UNK]": 33,
-  "a": 17,
-  "b": 14,
-  "c": 26,
-  "d": 12,
-  "e": 11,
-  "f": 5,
+  "a": 30,
+  "b": 18,
+  "c": 2,
+  "d": 27,
+  "e": 26,
+  "f": 16,
   "g": 19,
-  "h": 32,
-  "i": 6,
+  "h": 7,
+  "i": 11,
   "j": 31,
-  "k": 28,
+  "k": 22,
   "l": 23,
-  "m": 2,
-  "n": 29,
-  "o": 21,
-  "p": 18,
-  "q": 4,
-  "r": 8,
-  "s": 15,
-  "t": 16,
-  "u": 9,
-  "v": 20,
-  "w": 0,
-  "x": 13,
-  "y": 30,
-  "z": 24,
-  "|": 22
+  "m": 0,
+  "n": 32,
+  "o": 12,
+  "p": 17,
+  "q": 8,
+  "r": 1,
+  "s": 9,
+  "t": 10,
+  "u": 5,
+  "v": 3,
+  "w": 25,
+  "x": 28,
+  "y": 4,
+  "z": 13,
+  "|": 15
 }